Springer.Biostatistics.For.Radiologists
Total Page:16
File Type:pdf, Size:1020Kb
Biostatistics for Radiologists Planning, Performing, and Writing a Radiologic Study Francesco Sardanelli • Giovanni Di Leo Biostatistics for Radiologists Planning, Performing, and Writing a Radiologic Study Foreword by Adrian K. Dixon FRANCESCO SARDANELLI, MD GIOVANNI DI LEO, Dr.Sci. Associate Professor of Radiology Researcher University of Milan School of Medicine Radiology Unit Department of Medical and Surgical Sciences IRCCS Policlinico San Donato Director of Radiology Unit San Donato Milanese, Italy IRCCS Policlinico San Donato e-mail: [email protected] San Donato Milanese, Italy e-mail: [email protected] Originally published as: Biostatistica in Radiologia Progettare, realizzare e scrivere un lavoro scientifico radiologico Francesco Sardanelli, Giovanni Di Leo © Springer-Verlag Italia 2008 All rights reserved Library of Congress Control Number: 2008938920 ISBN 978-88-470-1132-8 Springer Milan Berlin Heidelberg New York e-ISBN 978-88-470-1133-5 Springer is a part of Springer Science+Business Media springer.com © Springer-Verlag Italia 2009 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illu- strations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the Italian Copyright Law in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the Italian Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Product liability: The publishers cannot guarantee the accuracy of any information about dosage and application contained in this book. In every individual case the user must check such information by consulting the relevant literature. Typesetting: Ferrari – Studio editoriale, Cologno Monzese (MI), Italy Printing and binding: Printer Trento S.r.l., Trento, Italy Printed in Italy Springer-Verlag Italia S.r.l., Via Decembrio 28, I-20137 Milan, Italy To Genny, Francesca, and Federica F.S. To my father, who taught me how to distinguish significant from nonsignificant things G.D.L. Foreword It is a pleasure to write the foreword to this excellent book on the evidence behind radiological investigations and biostatistics in radiology. This is an area which is not widely appreciated by radiologists and it would be an invaluable book for those in training who should become fully versed about terminology such as technical performance, diagnostic performance, diagnostic impact, therapeutic impact, patient impact, patient outcomes and societal impact. They should also know that the widely (and often erroneously) used term ‘accuracy’ may not be the best assessment! This very readable book should not only be useful for radiologists but also administrators who are now beginning to realise that an effective imaging department underpins all high quality cost-efficient modern medicine. The his- tory and development of evidence-based radiology (from Fryback and Thornbury’s original paper right up to date with recent contributions from Hollingworth, Hunink, Jarvik and Malone) is very well presented. Lorenzo Mannelli, an excellent Italian Radiologist working in Cambridge, says of the Italian Edition: “The book is easy to read and the short paragraph titles on the side of the pages make it easy to use for future reviewing of “hot topics” when needed. All the examples and vocabulary are from the radiolog- ical world, making the statistics easier to understand. Although after reading this book, you will still need statistical advice, at least you will be able to understand what the statistician is speaking about! The final chapter on impact factors is interesting and helps the reader to understand the dynamics of journals. I definitely recommend this book as an easy reading for residents in radiology”. I am sure that this new English language edition will fill a very major void in the radiological literature. The authors have done us a great service. Cambridge, UK, October 2008 Adrian K. Dixon, MD Preface to the Italian Edition Better to light a candle than to curse the darkness. CONFUCIUS Science deals with discovery but also with communication. It’s hard to say you have an idea if you are not able to evoke the same idea in the mind of your listener. MARCUS DU SATOY For many years “Biostatistics for Radiologists” was an unrealized dream of the senior author. Since that dream is now coming true, I have taken on the task of writing this preface, which offers the opportunity for an appraisal of the years leading up to the genesis of this book. I hope it can be useful to young col- leagues who intend to devote themselves to radiologic research. More than twenty-five years ago, I was a resident at the Postgraduate School in Radiodiagnostics of the University of Genoa, directed by Professor Luigi Oliva. My supervisor was Professor Giorgio Cittadini, Director of the Chair “R” of Radiology of the University. He was the chairman of my medical grad- uation thesis, entitled “Colonic hypotonic effect of fenoverine and hyoscine N-butyl bromide: analysis of variance with nonparametric tests”. Already then there was an attention to statistical methods predicting the events of my future before me. In 1984, after several years mainly dedicated to gastrointestinal double- contrast studies, I was included in a small team made up of physicians, physicists, and engineers who had the good fortune to work on one of the first magnetic resonance imaging scanners installed in Italy. It was a proto- type with a resistive magnet operating at only 0.15 T. For the best use of this new diagnostic technology, the radiologist had to understand the NMR phe- nomenon, the radiofrequency pulse sequences, and the role of the field gra- dients which generate the images. At that time, physicists were teaching magnetic resonance in courses and congresses using formal demonstrations based on Bloch’s equations, combining both classic and quantum models. These lessons were very hard to follow. Only when formulas and equations were translated into a different language, positively associated with the clin- ical meaning of the images, did the attending radiologists open their eyes and grasp the practical sense of that theory. For the first time I understood that scientific communication was a crucial process requiring intellect, fan- tasy, and creativity. X Preface to the Italian Edition In the same year, I was involved as author in a small paper entitled “Sensitivity, specificity, overall accuracy. What is the meaning of these three words commonly used in scientific radiologic language?”1. This paper was the result of an interesting discussion which had begun in Cittadini’s room late one evening and lasted for two-three hours. The topic – how to quantify diagnostic performance – seemed highly intriguing to me. I promised myself to gain a deeper insight into the matter. It was a new world waiting to be explored: how to evaluate the uncertainty intrinsic to biologic phenomena and measurements and, as a consequence, medical diagnosis. At the time I was only a resident cooperating towards writing an article, but I began to add some substance to the immaterial dream of writing a book. The chapter dedicated to “Indices of Diagnostic Performance” included in the Italian textbook “Diagnostic Imaging and Radiotherapy”, recently published in its sixth edition, would be the embry- onic stage of “Biostatistics for Radiologists”. Some years later, in 1987, I became a staff radiologist of the Chair R of Radiology at the Genoa University and San Martino Hospital. I began to add a planned research activity to the clinical routine. The areas of research involv- ing high-level cooperation with clinicians produced the most interesting results. This was the case with Giuseppe Molinari (cardiologist at the Genoa University) and Giuseppe Canavese, breast surgeon at the Genoa Cancer Research Institute. In this period I submitted my first manuscripts to peer- reviewed journals. Immediately, I understood that good technical knowledge and updated clinical experience are not enough for writing an article enough good to be accepted for publication. The crux of the matter is given by study design, data presentation and analysis, and, in particular, statistical methods which demonstrate the significance of the results. Around this time I began to interact with statisticians. Once again communi- cation was stifled. Radiologists were on one side of a wall and statisticians on the other, as it had been earlier with the physicists for magnetic resonance. In my personal experience, however, this wall crumbled thanks to research con- duced on multiple sclerosis. In this field, magnetic resonance imaging was playing an increasingly important role. My relations with Gianluigi Mancardi (Department of Neurology, University of Genoa) and Paolo Bruzzi (Clinical Epidemiology, Genoa Cancer Research Institute) were a turning point. Each of us wanted to learn what the other two colleagues already knew and would spend hours and hours to reach… understanding. At the same time, another apparently impersonal factor was in action – the reviewers analyzing the manuscripts I submitted to the journals. Their criti- cisms were sometimes very harsh. However, the higher the rank of the jour- nal, the greater the knowledge in methodology I could obtain by interacting with the reviewers, even though the manuscript was rejected. This was anoth- er way I begun to accumulate a limited know-how in biostatistics and research methodology applied to radiology. While I was (and still am) learn- 1 Sardanelli F, Garlaschi G, Cittadini G (1984) Sensibilità, specificità, accuratezza diagnostica. Quale significato attribuire a queste tre parole così spesso usate nel linguaggio scientifico radio- logico? Il Radiologo 23:58-59.